In the course of several drafts of this paper I have incurred a number of intellectual debts. Helpful comments on earlier drafts were received from Michael Edelstein, James Ferguson, Carter Goodrich, Frank Lewis, Peter McClelland, Donald McCloskey, Lloyd Mercer, Sherwin Rosen, Richard Thaler, and Harold D. Woodman, and from the members of the University Seminar in Economic History at Columbia University. I have also benefited from conversations with Lance Davis and Douglass North (and from their Institutional Change and American Economic Growth), as well as from many discussions on related issues with Robert W. Fogel. Detailed comments on style and substance by Paul David and the editor were most useful in the preparation of the final draft.

¹ For the most recent such burial announcement, see Lampard, Eric E., “United States: Periodical Literature,” Economic History Review, XXIII (1970), 411–16Google Scholar. Lampard does have enough insight into the academic mind to add a skeptical “but we shall see,” clearly expecting attempts to keep the corpse alive.

² The general preoccupation here will be with land grants to the transcontinental railroad systems. The grants from the federal government can be divided roughly into three categories: aids to transcontinental systems, to midwestern regional railroads, and to southern Reconstruction railroads. The shares of total land granted were approximately 77, 15, and 8 percent, respectively, justifying the emphasis on the former. However, the revenues of the midwestern regional railroads (in the undiscounted current dollar accumulations shown in the Public Aids volume) were almost 2½ times larger per acre than those of the transcontinentals, and amount to 31 percent of the total of net proceeds shown, as against 66 percent for the transcontinentals. Allowing for the difference in timing of sales, the importance of the regional land grants vis-à-vis the transcontinentals is even larger. U.S. Office of Federal Coordinator of Transportation, Public Aids to Transportation (Washington: G.P.O., 1940), II, pp. 107–11.Google Scholar,

³ Within the economic development literature this problem has been dealt with in Tibor Scitovsky, “Two Concepts of External Economies,” Journal of Political Economy, LXII (1954), 143–51Google Scholar. The central problem, one of possible failure in the signals given by the price system, is attributed to the lumpiness of the investment project, not to technical externalities in the customary sense of that term.

⁴ For this reason the problem of investment decisions in the presence of “building ahead of demand” is not quite the same as that discussed in Yoram Barzel, “Investment, Scale, and Growth,” Journal of Political Economy, LXXIX (1971), 214–31Google Scholar.

⁵ As Nathan Miller points out, the toll rates charged by the Erie Canal yielded a profit from operations, and provided a development fund for the state. Nathan Miller, The Enterprise of a Free Feople: Aspects of Economic Development in New York State During the Canal Period, 1792–1838 (Ithaca: Cornell University Press, 1962Google Scholar). “Building ahead of demand” seems more a logical statement—that a particular area could not be settled without a transport form—than one which should be chronologically interpreted. However implications for the governmental role do follow depending on whether the tracks precede or follow settlement. In the latter case, investors, already owning land, would be willing to finance the railroad to capture external effects on land values. This, indeed, is a typical pattern for the U.S. as well as for England. In the former case the problems of forming a coalition may be more difficult and thus require intervention by a higher level of government.

⁶ See Fishlow, Albert, American Railroads and the Transformation of the Ante- Bellum Economy (Cambridge: Harvard University Press, 1965), ch. vi.Google Scholar It has been suggested (by Harold D. Woodman) that this encouragement is what Jenks had in mind when he referred to the “railroad as an idea.” Leland Hamilton Jenks, “Railroads as an Economic Force in American Development,” THE JOURNAL OF ECONOMIC HISTORY, IV (1944), 1–20Google Scholar.

⁷ We shall restrict the discussion of decreasing costs to those instances based upon economies of scale internal to the firm. The case of the decreasing cost industry, with all firms having increasing costs, is thus excluded as not being part of the historical debate.

⁸ Another explanation of some cases of government construction and ownership has been provided in Rae, John Bell, “Federal Land Grants in Aid of Canals,” THE JOURNAL OF ECONOMIC HISTORY, IV (1944), 167–77CrossRefGoogle Scholar. Early discussion of aids to canals raised the prospect that they would be so profitable that all taxpayers should be permitted to benefit, and not just a select few. Private ownership, with an auctioning off of the franchise rights, could have achieved these results.

⁹ Mercer, Lloyd J., “Land Grants to American Railroads: Social Cost or Social Benefit?”, Business History Review, XLIII (1969), 134–51CrossRefGoogle Scholar, and Mercer, Lloyd J., “Rates of Return for Land-Grant Railroads: The Central Pacific System,” THE JOURNAL OF ECONOMIC HISTORY, XXX (1970), 602–26CrossRefGoogle Scholar.

¹⁰ This is a necessary, but not a sufficient, condition. For a discussion of this argument see Kemp, Murray C., “The Mill-Bastable Infant-Industry Dogma,” Journal of Political Economy, LXVIII (1960), 65–67CrossRefGoogle Scholar.

¹¹ For example, the Union Pacific legislation cited “postal, military, and other purposes” as a rationale for government aid. (Option demand refers to the desire to have the service available for potential future use, even if there is no demand at present, as could be claimed for the case of future military usefulness. National pride from a large, modern railroad network is another benefit which would not be captured by private pricing policy of the customary sort.)

¹² If paid annually, the subsidy would continue over the life of the investment. However, it could be paid in a lump sum of sufficient present value and with sufficient regulatory constraints to induce proper behavior over this lifetime. Regulation by itself could be used to eliminate monopoly profits, but this would not provide the socially optimum output.

¹³ See the classic article by Henderson, Alexander M., “The Pricing of Public Utility Undertakings,” Manchester School of Economics and Social Studies, XV (1947), 223–50CrossRefGoogle Scholar. For a recent approach that fits in the social decision-making process with the pricing discussion, see Buchanan, James M., “A Public Choice Approach to Public Utility Pricing,” Public Choice, V (1968), 1–17CrossRefGoogle Scholar.

¹⁴ The usual assumption is that, within any small area, the demand curve faced by producers for export is perfectly elastic at the “world” price less the costs of transport, and the supply of imports is also perfectly elastic, so that lowered transport costs do not affect the “world” price.

¹⁶ It is assumed that the railroad keeps the implicit contract concerning rates that it makes when selling the land. Actually, it would pay for the railroad to sell at a promise of low rates, but, once the land is sold for the higher price, to raise freight rates to the monopoly level. Note that unexpected rate wars could then provide landowners with windfall profits.

¹⁶ Land grants, with the alternate acreage provision, might still allow for the correct output, with taxpayers obtaining part of the “benefits.” There is a problem in evaluating the railroad pricing policy to maximize profits when the grant is for only one-half the adjacent land and when price discrimination is precluded. The output in this case, however, should be closer to the optimum output than it would have been in the absence of the grant.

¹⁷ As will be discussed in more detail below, the land-grant policy would make the railroad anxious to accelerate settlement, but that does not mean that it would be willing to part with land at less than its full value for a given fare structure. There would be no advantage to the railroad in foregoing those potential profits, nor is there one for society unless there is another decreasing cost component involved.

¹⁸ These decreasing cost considerations raise several issues for the interpretation of state failures to cover costs in earlier years, particularly in the canal era. There was no need for canals to be paying ventures, since some subsidy to cover fixed costs would have been desired, at least for certain pricing policies, and there are circumstances under which financial failure would have been the appropriate policy. In the analysis of canals undertaken by Harvey Segal the freight-rate used per ton-mile is less than half the resource costs per ton-mile, consistent with (among many other things) declining average costs. Harvey H. Segal, “Canals and Economic Development,” in Goodrich, Carter, et al. , Canals and American Economic Development (New York: Columbia University Press, 1961), pp. 216–48Google Scholar.

¹⁹ The imperfection could also have taken the form of a savings constraint, with the ordering of investments by social profitability differing from that by private profitability. Then a government subsidy could be necessary to provide the appropriate ordering by social profitability. For the nineteenth-century United States, with international capital inflows, this case seems of limited interest.

²⁰ The Public Aids data indicate that the use of government aids was more a function of capital market considerations than of externalities. A large portion of aids was not taxpayer financed, but consisted of a swap of government securities (federal, state, and local) for private securities. For example, of the total loans and subscriptions described, almost 90 percent entailed government payment in securities rather than in cash or in kind. Of the much smaller total of outright contributions, over 40 percent were made through securities, not cash. (The use of securities as opposed to cash was more frequent on the state than on the local level.) The same reliance on security issue rather than taxpayer finance was characteristic of the canal era as well. Harvey H. Segal, “Cycles of Canal Construction,” in Goodrich, Canals, pp. 169–215. To evaluate fully the costs of such policies, therefore, it is necessary to know more about the details of canal and railroad debt payments to governmental units.

²¹ The former method is seen in New Jersey's aid to canal construction, discussed in H. Jerome Cranmer, “Improvements Without Public Funds: The New Jersey Canals,” in Goodrich, Canals, pp. 115–66. The latter is found in Canadian land-grant policy, which left the railroad greater flexibility in choosing land than occurred in the United States, permitting land to be selected which was far removed from the location of construction.

²² In the case of the transcontinental and other land-grant railroads, the issuance of bonds backed by the grant was delayed, often until almost a decade after the grant was bestowed. For example, the Union Pacific did not issue land-grant bonds until 1869, and the Southern Pacific until 1871. Daggett, Stuart, Chapters on the History of the Southern Pacific (New York: Ronald Press, 1922)Google Scholar, and Fogel, Robert W., The Union Pacific Railroad: A Case in Premature Enterprise (Baltimore: Johns Hopkins Press, 1960Google Scholar).

²³ See Henry, Robert S., et al. , “The Railroad Land Grant Legend in American History Texts,” Mississippi Valley Historical Review, XXXII (1945–1946), 171–94CrossRefGoogle Scholar and 557–76.

²⁴ Mercer, “Land Grants,” p. 140.

²⁵ Presumably if purchased in small parcels, and without anyone else being aware of the purpose of such purchases.

²⁶ Vide the legislative battles for rights to franchises.

²⁷ In that event the private rate of return, including income from land revenues, would have been infinite when measured from the date the subsidy was received.

²⁸ See. Mercer, “Rates of Return,” and Cochran, Thomas C., “Land Grants and Railroad Entrepreneurship,” THE JOURNAL OF ECONOMIC HISTORY, X (1950), 53–67Google Scholar.

²⁹ The extrapolation of a $1 subsidy value per acre would make federal land grants equal to about one-fifth of the government investment in railroads calculated by Goodrich. Goodrich's estimate of the proportion of government investment to total investment, with his suggested addition of the value of land to the total of other aids, is misleading. The latter total does not really provide an estimate of the net amount of government subsidy, but rather of the total of government cash and security aids. Given the importance of the swapping of private for government securities, the proportion of taxpayer subsidy represented by land grants would be higher than the average subsidy implicit in the other categories of aid. See Goodrich, Carter, “Internal Improvements Reconsidered,” THE JOURNAL OF ECONOMIC HISTORY, XXX (1970), 289–311CrossRefGoogle Scholar.

³⁰ Thus, relative to the size of the federal budget in the year granted, the Illinois Central grant was the most important of the land grants.

³¹ For example, the Union Pacific had sold only 60 percent of its land by 1890, and the Northern Pacific, by far the largest recipient of land, did not sell one-half of its grant until early in the twentieth century. Twenty years after the grant less than 10 percent had been sold. While the Illinois Central was atypical in both selling price and pace of sale, most of the other midwestem lines did achieve sales of over one-half the land in their grants within about 20 years. See Gates, Paul W., The Illinois Central Railroad and Its Colonization Work (Cambridge: Harvard University Press, 1934CrossRefGoogle Scholar); Overton, Richard C., Burlington West: A Colonization History of the Burlington Railroad (Cambridge: Harvard University Press, 1941CrossRefGoogle Scholar); Poor's Manual of the Railroads of the United States (New York: Annual 1868Google Scholar); and Waters, L. L., Steel Trails to Santa Fe (Lawrence: University of Kansas Press, 1950Google Scholar). It remains a mystery why the transcontinentals took so long to sell their land. The lag is inconsistent with the perfect foresight assumption.

³² On the general question of the speed of sales by landowners, see Robert W. Fogel and Jack Rutner, “The Efficiency Effects of Federal Land Policy, 1850–1900: A Report of Some Provisional Findings,” in Aydelotte, William et al. (eds.), Dimensions of Quantitative Research in History (Princeton: Princeton University Press, 1972Google Scholar). Again, however, it is unclear exactly why the railroads did take so long in selling their lands, even if the sales were made at a more rapid pace than would have occurred if the government had been the seller.

³³ Mercer, “Rates of Return.” This is Mercer's second test, which he considers to be conclusive. The first criterion, relating to the short-fall of the private rate of return to investors below the market rate, is, however, important for determining both the desirability of a subsidy policy and the size of the optimum subsidy. Mercer does point out that his verdict “rests on the further conclusion that land grant policy hastened construction of the system.” The support for this judgment rests on a statement made by Robert Edgar Riegel, which places the acceleration at ten to fifteen years. Riegel, Robert E., The Story of the Western Railroads (New York: Macmillan, 1926Google Scholar). Riegel's estimate is frequently cited, but is clearly more an impression (albeit an educated one) than the specific result of the testing of any hypothesis. Not everyone, of course, accepts this possible acceleration of railroad construction as necessarily a desirable condition for the national economy. See, for example, Cleveland, Frederick A. and Powell, Fred W., Railroad Promotion and Capitalization in the United States (New York: Longmans, Green, 1909), p. 253Google Scholar; and Cochran, Thomas C., “Did the Civil War Retard Industrialization?”, Mississippi Valley Historical Review, XLVIII (1961), 197–210CrossRefGoogle Scholar, where it is argued that the economy would have been better-off if construction had been delayed.

³⁴ See Peter D. McClelland, “Social Rates of Return on American Railroads in the Nineteenth Century,” (Unpublished, Discussion Paper 188, Harvard Institute of Economic Research, April 1971).

³⁵ This would happen if land values were high without the railroads and the railroads added little to any intertemporal increase in value, or if the aid took the form of other profitable assets, permitting entrepreneurs to obtain private profits even though the railroads provided no net benefits for society.

³⁶ If government expenditures were held constant, in essence the transaction would amount to an increased tax burden upon the citizens to pay a certain sum to the railroad investors, and since taxation was not lump-sum, some distortions could occur.

³⁷ For an application of this concept to a discussion of the Canadian Pacific, see George, Peter J., “Rates of Return in Railway Investment and Implications for Government Subsidization of the Canadian Pacific Railway: Some Preliminary Results,” Canadian Journal of Economics, I (1968), 740–62CrossRefGoogle Scholar. Of course, using the ex post data developed by Mercer for the two American transcontinental, systems, the “optimum subsidy” would have been zero.

³⁸ Robert W. Fogel, in evaluating alternative methods of providing for the construction of the Union Pacific, analyzes one important aspect of the issue of social benefits and costs of land-grant subsidies. He considers how construction could have been undertaken at lowest social cost, and concludes that under specified conditions it would have been through government construction. See Fogel, Union Pacific, ch. iv.

³⁹ For a discussion of the dispute between the settlers and the railroads on this count see Gates, Paul W., History of Public Land Law Development (Washington: G.P.O., 1968), chs. xivGoogle Scholar, xvi.

⁴⁰ The issue of the extent to which the railroad deferred taking title to land to avoid, taxation has recently been studied in the states of Kansas and Nebraska. The author “raises the possibility that the railroads may have been taxed appropriately, or even excessively, in these states. Decker, Leslie, Railroads, Lands and Politics: The Taxation of Railroad Land Grants, 1864–1897 (Providence: Brown University Press, 1964Google Scholar).

⁴¹ See, for example, Swierenga, Robert, Pioneers and Profits: Land Speculation on the Iowa Frontier (Ames: Iowa State University Press, 1968Google Scholar). The Homestead Act provisions were apparently not important in the land-grant states, and there were restrictions imposed on their usage within the land-grant limits.

⁴² The government imposition of a double-minimum price on land within’ the grant limits, whatever its political appeal, could only mean a slower rate of settlement. If all the government sales were at the legal minimum price, this double-minimum would also have reduced the extent of redistribution from taxpayers to land purchasers. For a convincing argument that the government did not achieve its hopedfor financial ends, see Gates, Paul W., “The Railroad Land-Grant Legend,” THE JOURNAL OF ECONOMIC HISTORY, XIV (1954), 143–46CrossRefGoogle Scholar.

⁴³ See, in particular, Ellis, David M., “Railroad Land Grant Rates, 1850–1945,” Journal of Land and Public Utility Economics, XXI (1945), 207–22CrossRefGoogle Scholar. This issue is not considered by Mercer, who does not distinguish the government from other shippers. This is consistent with his procedure of regarding the railroad investors within the group for whom social benefits and costs are measured, and thus putting aside the division between taxpayers and investors.

⁴⁴ If the railroads were covering total costs on other traffic before the generation of government traffic, there would be no need to alter the other rates, so no additional social costs would be imposed.